Robust space-time coding for fading and single-path MIMO wireless channels

The evolution of wireless communication systems toward high data throughput applications received a significant boost with the discovery that the capacity on a MIMO (Multiple-Input, Multiple-Output) fading channel grows in proportion to the array size. Since that time, researchers have sought space-time coding techniques to exploit the immense potential capacity of MIMO wireless systems. However, progress has not come easily. Codes are difficult to design and unreasonably cumbersome to decode for large numbers of transmit antennas, where the potential capacity is most attractive. And, all along, this work is bound to the assumption that the channel is rich scattering, which may not be the case for practical deployment of such systems. This dissertation explores the validity of modeling the channel as rich scattering and proposes a simple, scalable space-time code with robust error performance over a variety of channels. A physical channel model is developed that employs linearly-polarized antenna elements in an artificial scattering environment. Results show both greater intra-array correlation than predicted by the rich scattering model and the possible benefit of exploiting polarization diversity. On error performance curves the higher correlation of the physical channel path gains results in a noticeable decrease in diversity slope. In an effort to address the concern that deployed wireless units will encounter very different (and thus not necessarily amenable to the rich scattering model) physical environments, a space-time code is presented that provides robust error performance over the range of channels from the non-fading single-path to rich scattering. This design overlays trellis coded modulation on a block orthogonal design and outperforms some basic space-time block and trellis codes on the family of Rician channels (spanning the range from non-fading single-path to rich scattering) as well as the physical channel model. In each case, the overlay design performs within about 5 dB of the corresponding Fano bound. The attractiveness of this design owes to its adaptability to different physical environments, its simple implementation, and its scalability to larger array sizes.